Half right is the most dangerous place to be when it comes to anything involving technology.

I think about this every time I see something cobbled together in a weekend to which the author says “This was 100% vibe coded, I could have never done this so quickly on my own!”

With rare exception, I look at the end result and think “I definitely could have done that in a weekend too, coded the old fashioned way. And not only that, I’d understand how every part of what I built worked, and what tradeoffs I made while working on it to get it done sooner than later”

But that too, is only half right. I could say that confidently about anything I’ve already done before, using the tools I’m familiar with. I could say that if I controlled the scope, and the tolerance levels around tech debt.

Throw in even the smallest complicating factor, and I’d need to spike to get a real answer. It’s possible that my entire plans could be thrown out the window by some random system dependency that’s very difficult to install, even if that’s buried deep down six levels away from the problem I’m actually trying to solve.

Vibe coding might route around that with a random flag to disable some newer security checks that otherwise would prevent the whole program from running. I might do that too, if I knew the risks involved and made an informed choice.

The vibe coder who copy pasted the cryptic error back into their LLM without even reading it? They might never know, or care.

What did GenAI prove about the software development world as it sits right now? That it’s an enormous mess and that a large percentage of the population would rather sweep that under the rug rather than fix the issues at the roots.

That’s the danger of being half right.

It’s time to stop pretending that “normal life” exists in a world that’s burning.

The amount of focus and energy required by any sort of skilled labor is higher than what most folks have capacity for at the moment. So many are tired, sick, scared, enraged, endebted, and <insert other bad things here>.

Money can be used as a salve by those who have it and those who can still borrow it, but the cost of convenience, comfort, and safety is trending to infinity. Most everyone else is watching every dollar they make burn alongside the world, simply to get through the day.

Failure to adapt is where most suffering comes from. But it’s not clear that any of us know how to adapt to the world we’re living in.

So it seems worthwhile to get back to basics.

To do that I’ve been looking at Maslow’s Hierarchy of Needs, which states the semi-obvious… figuring out how to pay for and acquire the food on your plate always comes before building capacity to solve the more complex problems of life.

Hofstadter’s Law: It always takes longer than you expect, even when you take into account Hofstadter’s Law.

This statement from the author of Gödel, Escher, Bach is one of those things that has rattled around in my head for decades because of how true it is. And yet, it is one of those observations that feels like there’s nothing that can be done about it… and so it’s a bit like saying “I just lost the game” every time you think of The Game.

At least that’s what I thought, until I saw a Mastodon post from Robert Roskam that opened up a new line of thinking for me. In particular, he suggests that “The solution isn't better estimation; it's shorter feedback loops.”

I’ve been mulling that over since I saw the post the other day.

My gut reaction is to say “Aha! Feedback loops are indeed at the root of all this… but is shorter always the answer?”

That’s my systems thinking brain kicking in… feedback loops are indeed often too slow in most human activities. But on the other hand, oscillations and vicious cycles can be caused by feedback loops that are either too short or too frequent… so there’s a balance to be found there that is probably quite context dependent.

And so perhaps it’s worth deliberately blurring things a bit to leave room for more reflection, while still serving as a navigational aid of sorts, with this rephrasing:

“The solution isn't better estimation; it's better feedback.”

This leaves it up to the individual to define more precisely what better feedback means.

But it still calls out that the cause of delays (particularly unanticipated delays (and particularly unanticipated delays despite a lot of analysis being put in to determine the likely sources of delays (and particularly when that happens even when you’re aware of the phenomena ))) — is probably in whole or in part because the right information didn’t reach the right place at the right time with the right level of detail to allow for swift course corrections.

This is a new thought for me! Or at least a new angle on a pattern that shows up in a thousand different forms.

So I’m thankful to Robert for stirring it up. :)